the L2-norm ball in case of a mixed-integer problem #269
Conversation
Codecov Report❌ Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #269 +/- ##
==========================================
- Coverage 86.23% 84.83% -1.40%
==========================================
Files 17 22 +5
Lines 1482 2691 +1209
==========================================
+ Hits 1278 2283 +1005
- Misses 204 408 +204 ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
|
only 2normball here |
src/polytope_blmos.jl
Outdated
| count = 0 | ||
| for idx in 1:length(int_vars) | ||
| if !(lb[idx] <= 0 <= ub[idx]) | ||
| count = count + 1 |
| if !(lb[idx] <= 0 <= ub[idx]) | ||
| count = count + 1 | ||
| end | ||
| if count > 1 |
There was a problem hiding this comment.
A comment would be helpful here, the code is not very clear.
There was a problem hiding this comment.
So we are just tackling the case of radius=1? Then, this has to be more clear in the function definition.
src/polytope_blmos.jl
Outdated
| BLMO denotes the L2normBall, It is unit ball which means R = 1 | ||
| """ | ||
|
|
||
| struct L2normBallBLMO <: FrankWolfe.LinearMinimizationOracle end |
There was a problem hiding this comment.
I don't think we need an extra struct here.
src/polytope_blmos.jl
Outdated
|
|
||
| struct L2normBallBLMO <: FrankWolfe.LinearMinimizationOracle end | ||
|
|
||
| function bounded_compute_extreme_point(blmo::L2normBallBLMO, d, lb, ub, int_vars; kwargs...) |
There was a problem hiding this comment.
Write instead
function bounded_compute_extreme_point(blmo::FrankWolfe.LpNormBallLMO{T,2}, d, lb, ub, int_vars; kwargs...)
...
end
And the same for the other two functions here.
| end | ||
|
|
||
| # Test for L2normBallBLMO | ||
| @testset "L2normBall BLMO continuous" begin |
There was a problem hiding this comment.
You are not testing the continuous case here, though.
There was a problem hiding this comment.
Now it test the continuous case
There was a problem hiding this comment.
You still have num_int=5. So it is still an integer problem.
There was a problem hiding this comment.
But for the optimum, the integer entries (dimension) are all 0, the values are all in continuous entries.
test/LMO_test.jl
Outdated
| lower_bounds = fill(-2.0, num_int) | ||
| upper_bounds = fill(2.0, num_int) |
There was a problem hiding this comment.
Directly, fill with 1.0 and -1.0 directly.
| # Generate a solution inside the L2 ball (||x|| <= 1) | ||
| x_sol = randn(rng, n) | ||
| x_sol = x_sol ./ (norm(x_sol) * 1.5) | ||
|
|
||
| # Round some coordinates to integers for testing | ||
| num_int = 5 | ||
| int_indices = sort(rand(rng, 1:n, num_int)) | ||
| for idx in int_indices | ||
| x_sol[idx] = 0 | ||
| end |
There was a problem hiding this comment.
How can you guarantee that x_sol will stay inside the ball? If you cannot, the tests later will fail even if we find the correct solution.
There was a problem hiding this comment.
x_sol = x_sol ./ (norm(x_sol) * 1.5) here guarantee that x_sol will stay inside the ball
test/LMO_test.jl
Outdated
| @test isapprox(f(x), f(result[:raw_solution]), atol=1e-6, rtol=1e-3) | ||
| end | ||
|
|
||
| @testset "L2normBall BLMO integer 1" begin |
There was a problem hiding this comment.
There is not difference between the tests? One is sufficient.
LJS42
changed the title
| """ | ||
| 2normBallBLMO() | ||
| BLMO denotes the L2normBall, It is unit ball which means R = 1 | ||
| """ | ||
|
|
There was a problem hiding this comment.
Please add some documentation here.
There was a problem hiding this comment.
Yes we are just tackling the case of radius=1. I am still thinking and working on the general case.
| if !(lb[idx] <= 0 <= ub[idx]) | ||
| count = count + 1 | ||
| end | ||
| if count > 1 |
There was a problem hiding this comment.
So we are just tackling the case of radius=1? Then, this has to be more clear in the function definition.
| end | ||
|
|
||
| # Test for L2normBallBLMO | ||
| @testset "L2normBall BLMO continuous" begin |
There was a problem hiding this comment.
You still have num_int=5. So it is still an integer problem.
2 participants
No description provided.